The dream of watching a protein function in real time with near atomic resolution has been realized using time-resolved Laue crystallography. To further this capability, which we helped develop at the ESRF in Grenoble, France, we have launched a major effort to develop a picosecond time-resolved X-ray source at the Advanced Photon Source (APS) in Argonne, IL. In addition to our efforts to develop improved hardware for making these measurements, which is summarized in a separate report, we have also made further progress in developing software for analyzing these data. Though the effort to develop a stand-alone package capable of analyzing Laue data has proven to be much greater than envisioned at the outset, it represents a critical component in our research. This package, called TReX, continues to evolve, but now contains most of the features required to analyze time-resolved Laue diffraction data and generate molecular movies that unveil protein structure changes in real time. ? ? The first step in analyzing crystal diffraction data, whether acquired with monochromatic or polychromatic X-ray radiation, is the indexing of diffraction spots recorded on a two-dimensional detector. Robust auto-indexing algorithms have long existed for monochromatic diffraction images, but Laue diffraction images, which are generated with a polychromatic X-ray source, are not amenable to those methods. Consequently, the analysis of Laue diffraction data has generally proven to be a time-consuming, off-line process that has required a significant amount of face time in front of a computer. An approach for auto-indexing Laue diffraction images has been implemented in TReX; however, that method has occasionally failed when attempting to index a crystal that doesn't diffract as well as our best crystals. It would be preferable to employ an improved auto-indexing method in TReX. A major step forward in developing a robust real-time approach to analyzing Laue diffraction data was recently developed by Dr. Eric Henry, who pursued a new auto-indexing method based on zones. When the image center and distance between the sample and the detector are prescribed, spots on the detector plane can be mapped onto a locus of possible zone-vector directions, the consensus of which identifies zone-vectors suitable for determining the orientation of the crystal. This algorithm is undergoing final testing and will hopefully be implemented in TReX in the fall of 2007.? ? The X-ray diffraction from protein crystals generally degrades when exposed to high fluence laser pulses, and this degradation is often manifested as streaky diffraction spots. In effect, the crystals suffer from pump-induced mosaicity. Once the pump pulse energy is sufficient to photoexcite most chromophores within the X-ray probed volume in the crystal, its energy is also sufficient to induce a T-jump in the excited volume. The thermal expansion that ensues can lead to a small degree of curvature in the long-range order. Because the crystal acts as a monochromator, a strained crystal will diffract different X-ray wavelengths in slightly different directions, which will be manifested as streaky spots. Currently, no Laue data analysis packages account for the wavelength dependence of the streaky spots, which are treated as if all photons assigned to them share the same wavelength. Because the undulator spectrum is sharply peaked, accurate integration of spots that arise from wavelengths near the peak of the spectrum is quite challenging. One of our collaborators, Dr. Eric Henry, is currently working on this problem. Together, we have developed a strategy that accounts for the wavelength dispersion in the measured spots, and once implemented, should produce the most accurate structure factors available from our time-resolved Laue diffraction images. ? ? Most X-ray beamlines operate 24/7. Because the setup requirements for time-resolved studies are significant and the time allocated to our studies is limited, we generally pursue our experimental studies around the clock for several days straight before turning the beamline over to the next user. To make best use of this precious beam time, it is invaluable to have rapid feedback regarding whether the experiment is working well or not. To that end, our TReX software is designed to analyze diffraction images as fast as they are acquired. By analyzing and visualizing our data in real time, we are able to make better decisions regarding how to use our remaining beam time. This capability is especially valuable when trying to function while sleep-deprived. ? ? We developed a novel method for visualizing structural changes in proteins by overlapping color-coded electron density maps of the ground (magenta) and photolyzed (green) states. The magenta-to-green color gradient informs us about the direction of the atomic motion; where the atomic coordinates are stationary, the two colors blend to white. By stitching together a series of time-resolved snapshots into a movie, we are able to visualize molecular motion in real time with about 100-ps time resolution and near-atomic spatial resolution. This method provides a powerful way for assessing the correlated molecular motion in a qualitative fashion. However, this method is not quantitative. To quantify the time-dependent population and the time-dependent displacement of the atomic coordinates, detailed modeling of our data is required. In collaboration with the group of Prof. George Phillips at the Univ. of Wisconsin, we employed a difference-refinement approach to generate atomic models that were constrained by our picosecond time-resolved electron density maps. This approach has been extended to long-time studies of ligand dynamics in MbCO with and without pressurized Xe in the capillary. The presence of Xe in the Xe1 docking site blocks access to that site, which allowed CO to become sequestered in the Xe2 docking site. Prior to this study, we have never seen ligands passing through that site. Moreover, we see evidence for water leaking into the hydrophobic interior of the protein. By modeling the time-dependent electron densities in our movies, our collaborators are able to assess quantitatively the time-dependent occupancy of various ligands in various sites along the ligand migration pathway.
| Aranda 4th, Roman; Levin, Elena J; Schotte, Friedrich et al. (2006) Time-dependent atomic coordinates for the dissociation of carbon monoxide from myoglobin. Acta Crystallogr D Biol Crystallogr 62:776-83 |
| Bourgeois, Dominique; Vallone, Beatrice; Arcovito, Alessandro et al. (2006) Extended subnanosecond structural dynamics of myoglobin revealed by Laue crystallography. Proc Natl Acad Sci U S A 103:4924-9 |
